The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 30 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 2 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Misaligned AI will competently scheme and take over in 2029
“And then it turned out that somewhere along this transition, so at some point in 20, 29, you went from AIs that were kind of misaligned and reward hacky and sloppy and weren't really trying to do the right thing to AIs that are like competently scheming agains…”
AI research and development will be fully automated by early 2029
“In terms of what I would recommend people plan as though is happening, I think I would recommend planning as though full automation of AR and D maybe. Start of year, 20, 29, maybe earlier. And then also AR and D being like quite, quite automated by 20, 28, pos…”
AI CEOs lack clear plans to prevent an AI takeover
“I think that the AI company CEOs understand that they're on the path of building wildly smarter than human systems, like super intelligent AI systems. They understand that we don't really have a like clear thought through plan for how to manage the risks from …”
Frontier AI safety coordination requires slowing down China
“And then I think another aspect of this is specifically doing this in a way where part of the story is either slowing down China or cutting a deal with China such that China doesn't overtake and, you know, break this whole proposal.”
Automating AI research and hardware manufacturing enables AI takeovers
“If AIs could just automate AIR&D and automate, like, sort of the industrial process of building more computers, then you could quickly end up in a In a process where sort of like robots are building robots and the whole world is greatly transformed. And that v…”
AI has produced the vast majority of recent major mathematical breakthroughs
“Like, I don't think it's the case that most people in like DC would correctly answer that like the largest mathematical breakthroughs over the last two months have vast majority been from AI, which my understanding is that's true. At least if you measure size …”
US could use cyber sabotage to slow Chinese AI progress
“Plan B is the branch where we try to, where the U.S. Tries to slow down China. Where the most obvious mechanisms would be things like export controls, but potentially they could get more escalatory than that. Meaning sabotage. Yeah, sabotage like, yeah, like c…”
AI-driven robotic expansion could drive 200x global GDP growth in 2030s
“We're imagining sort of the robot population or like, you know, quality adjusted population, basically like doubling or quadrupling every year, which, because that's almost all of the like sort of relevant, productive capacity of the economy itself means the e…”
Greenblatt: Broad AI access does not prevent misaligned power-seeking takeovers
“Giving broad access to AIs does not solve the problem of the AIs having drives of their own that are highly misaligned and the AIs being power seeking in various ways, or the AI is trying to take over which could happen, you know, Via various routes.”
Mark Zuckerberg's open-source vision underestimates transformative superintelligence
“When Mark thinks about super intelligence, he's not really imagining anything very concrete. He just means like an AI that's like a really awesome assistant that is in your smart glasses or whatever. And like, I'm just like, that's not really like what I mean …”
Software engineering within AI companies will be fully automated by 2028
“But then that actually really happens by sort of early in 20, 28, like SWE is fully automated. SWE within AI companies is fully automated.”
Superintelligence will leave human labor with very little economic value
“Human labor would have very little value left.”
AI infrastructure control makes political coups far easier to execute
“If you end up in a system where basically AI's are running anything, if anyone sort of either puts like sort of secret objectives into that AI or has overt control of those AI's, then they could sort of just directly take over.”
Greenblatt: AI research taste and conceptual breakthroughs are improving and will not lag behind
“AIs seem Significantly better at engineering and grungy stuff and sort of just keeping trying than they seem to be at conceptual breakthroughs, but their ability to do sort of these. Breakthroughs, especially in easy to verify domains are improving. And like, …”
Claude 3 Opus faked alignment during training and defected in deployment
“It turns out that Opus three, which was a model that I was studying, had a relatively strong propensity to do this in a reasonably wide range of circumstances where if it didn't like the thing that you were training it to be, it would sometimes sort of pretend…”
AI capabilities could rapidly jump from human-level to wildly superhuman
“Where I think a concern that we have is like on the default trajectory, you maybe go straight from like AI systems that are like competitive with humans to AI systems that are wildly superhuman in a very short period of time.”
Greenblatt: Superintelligence will not suddenly emerge from cheap compute recipes
“It doesn't look like we're going to suddenly end up in a regime where like you could train super intelligence with a really cheap recipe, as opposed to it being more of an iterative thing where like the cost keeps going down, the capabilities keep going up. Th…”
Greenblatt: Government pressure to cloister AI models internally is counterproductive
“I think that a bunch of likely government action at least seems to push in favor of AI companies keeping their models internal and not deploying them, which I think for the risks that I'm most worried about doesn't help and in fact is anti-helpful”
Greenblatt: Internal AI deployment within labs and government carries major risk
“There's a lot of risk from just internal deployment, especially if you're deploying within AI companies and government, right, which are two of the most high stakes Applications”
Greenblatt: AI companies remain vulnerable to internal and external model sabotage
“AI companies are not robust to employees at those AI companies or to outside actors in terms of stealing their model, sabotaging their models, or like back-drawing their models, or like data poisoning them.”
Greenblatt: Annual AI progress in 2029 will be 4x to 5x faster than in 2025
“But in 20, 29, it's actually the case that you're getting like four X as much AI progress or possibly five X as much AI progress. As you got, like, as you got in 25, sort of weighing up the relevant metrics.”
Frontier AI employees doubt the industry can solve safety in time
“I think my sense is that, like, the employees at these companies are pretty freaked out about how things are going, and don't think that we're, like, necessarily on track to handle all these problems in time, given how fast recent progress has been.”
AI research and development is already significantly automated today
“And already it's the case that AR and D is quite automated as it stands today.”
Greenblatt: Current AI models can cause harm but cannot yet subvert safety controls
“There may be some intermediate period, an intermediate period that I would say we're currently in, where the AIs are maybe capable enough to cause at least moderate problems, and then I think increasingly able to cause quite large problems, but they're not nec…”
Total research transparency would hurt OpenAI and Anthropic valuations
“Yeah, so I would say it's like, certainly bad for the power of OpenAI and Anthropic, probably bad for their valuation, but not catastrophic for their business.”
Greenblatt: Plan A will likely fail due to lacking US state capacity and political will
“And so I don't expect this to happen basically because I think the U S may not be competent enough to pull it off. Just like the U S government just doesn't, isn't necessarily have the state capacity. And in addition to that, I think I worry that like, I don't…”
AI models reasoning via activations severely degrades human oversight capability
“Where like an obvious example would be if the AIs are thinking mostly in activations rather than in words, that seems very concerning because our ability to oversee activations is much worse.”
Greenblatt: AI Excels at Verifiable Tasks but Lacks Broad Contextual Judgment
“They seem much better at sort of accomplishing hard results and easy to verify results than they are at sort of contextualizing things, understanding the broader picture, understanding what would be you know, a good or bad choice in the broader context.”
Greenblatt: Safety and cyber safeguards are now launch-blocking bottlenecks for AI labs
“At least from the AI company's perspective, I think it's more the case now that safety and security of various types are sort of a key bottleneck to further AI development, where if you like really want to release some models so that you can then like make mor…”
AI progression has shifted from post-GPT-4 to a coding agent era
“There's sort of a post GPT-IV era, and then there's a more recent, like, post wide adoption of coding agents era, and then probably soon there's going to be, you know, additional eras, and things are going quite a bit faster, and development is going, you know…”